🆕 blog! “Removing "/Subtype /Watermark" images from a PDF using Linux”
Problem: I've received a PDF which has a large "watermark" obscuring every page.
Investigating: Opening the PDF in LibreOffice Draw allowed me to see that the watermark was a separate image floating above the others.
Manual Solution: Hit page down, select image, delete, repeat 500 times. …
👀 Read more: https://shkspr.mobi/blog/2026/01/removing-subtype-watermark-images-from-a-pdf-using-linux/
⸻
#LLM #pdf #python
@Edent given you have vibe coded a scriopt to remove the watermark can we also assume that the 'large watermark' is there to tell everyone that the contents of the pdf were originally generated by an LLM?
If so can't have viewers seeing that can we?
@marjolica eh? I've no idea what you're talking about.
It was a pre-print book with a publisher's watermark.
@Edent Hi! The watermarks are everything from /Artifact (this is actually the first argument to the BMC operator) to EMC. They don't have a length, but once you remove them you have to fix up the /Length on the enclosing content stream. However! Since you decompressed the stream, you can just set the / Length to 0, and any PDF viewer can figure it out.
@Edent @dhd6 @jleedev
I have done similar with Inkscape to remove watermarks and, worse, redaction blocks. It was some time ago & maybe only twice. Inkscape can be driven by script, Inkscape selects/operates on single page so I used pdftk to split doc out to single pages, ran the Inkscape script, and merged pages again with pdftk. Apols that I can’t dig out & share specifics.
@Edent @jleedev if you want to better understand the structure of a PDF you can use my software 😀 but it's strictly read only https://dhdaines.github.io/playa/latest/
@Edent Replacing the content with whitespace works just as well without having to fix up the /Length.
